[Biojava-dev] Bug when reading FASTA file with many DNA Sequences
Mapleson Daniel Dr (CMP)
D.Mapleson at uea.ac.uk
Wed Feb 9 16:22:58 UTC 2011
Thanks Scooter. That's great. This jar fixes the problem. The file is loaded really quickly and memory usage has decreased massively too.
We will need to process each (most) sequence(s), in the file at some stage, so the ability to free up memory containing sequences that aren't currently being processed/used will be useful going forward with our project. It's not urgent though. The main thing from my perspective is that we can actually run the program, which your fix allows us to do.
Thanks again for the quick turnaround! Much appreciated! :)
Best regards,
Dan
>-----Original Message-----
>From: Scooter Willis [mailto:HWillis at scripps.edu]
>Sent: Wednesday, February 09, 2011 3:50 PM
>To: Mapleson Daniel Dr (CMP); biojava-dev at lists.open-bio.org
>Subject: Re: [Biojava-dev] Bug when reading FASTA file with many DNA
>Sequences
>
>Dan
>
>I have attached a copy of my biojava3-core that has that method in it as
>well as other memory/speed optimizations I worked on. Sounds like that
>method(recently added) hasn't made its way into the current biojava3
>jars.
>You should see a dramatic reduction in memory if you only need to select
>a sub-set of sequences. Trying to load a 245MB fasta file does take lots
>of memory. If you plan on reading each sequence then you will eventually
>run into a memory problem as I am currently not freeing up the sequence
>data that is loaded lazily. My plan is to add some optimization
>hints/logic that the developer can control that every time you load a
>new sequence and use more memory I will internally free up sequence data
>that has been allocated. If you go back to a sequence that has had
>storage deallocated then I will simply reload it. This way you can work
>with very large sequence files at a genome scale without running out of
>memory or being forced to put in a database.
>
>Let me know if this works and if you need to analyze every sequence and
>will see if I can find some time to add in the lazyload memory
>management features.
>
>Thanks
>
>Scooter
>
>
>
>
>
>
>On 2/9/11 10:07 AM, "Mapleson Daniel Dr (CMP)" <D.Mapleson at uea.ac.uk>
>wrote:
>
>>I'm running Windows 7 Enterprise 64-bit and Java 1.6 64-bit. I double
>>checked I was using the 64-bit version of java at runtime using
>JConsole.
>> I'm also using biojava3, in case that makes a difference.
>>
>>I tried the FastaReaderHelper.readFastaDNASequence(File f) version of
>>the method, but I still haven't found the lazySequenceLoad version.
>>Same problem.
>>
>>Best regards,
>>Dan
>>
>>>-----Original Message-----
>>>From: Scooter Willis [mailto:HWillis at scripps.edu]
>>>Sent: Wednesday, February 09, 2011 2:48 PM
>>>To: Mapleson Daniel Dr (CMP); biojava-dev at lists.open-bio.org
>>>Subject: Re: [Biojava-dev] Bug when reading FASTA file with many DNA
>>>Sequences
>>>
>>>Dan
>>>
>>>I usually do 40MB DNA files with no problem. I will concat together
>>>and test a 245MB version. What operating system and version of Java?
>>>32bit or 64bit? The GC should be able to keep up in lazySequenceLoad
>>>mode.
>>>
>>>You need to use File because an inputstream doesn't provide the
>>>ability to random seek based on an offset as it could be an HTTP
>>>stream etc.
>>>
>>>Thanks
>>>
>>>Scooter
>>>
>>>On 2/9/11 9:38 AM, "Mapleson Daniel Dr (CMP)" <D.Mapleson at uea.ac.uk>
>>>wrote:
>>>
>>>>Hi Scooter,
>>>>
>>>>Thanks for the quick feedback.
>>>>
>>>>Unfortunately, the memory isn't the issue, I set my JVM to use 2500MB
>>>max
>>>>heap (the most I can get away with on my machine), and still
>>>encountered
>>>>the same problem. On a colleagues machine he has the max heap
>>>>essentially unbounded and he still gets the same error. It seems to
>>>>be something to do with the garbage collector removing temporary
>>>>items
>>>from
>>>>memory rather than max available memory.
>>>>http://stackoverflow.com/questions/1393486/what-means-the-error-
>>>message-ja
>>>>va-lang-outofmemoryerror-gc-overhead-limit-excee
>>>>
>>>>Also the FastaReaderHelper.readFastaDNASequence method ran into the
>>>same
>>>>problem. I passed in an input stream rather than a file but I don't
>>>>think that should cause the problem, should it? Also I couldn't find
>>>an
>>>>overloaded variant with the lazySequenceLoad signature. Maybe I'm
>>>using
>>>>an older version but I couldn't find it in the biojava3 API docs
>>>either.
>>>>
>>>>Any other ideas?
>>>>
>>>>Best regards,
>>>>Dan
>>>>
>>>>
>>>>>-----Original Message-----
>>>>>From: Scooter Willis [mailto:HWillis at scripps.edu]
>>>>>Sent: Wednesday, February 09, 2011 1:31 PM
>>>>>To: Mapleson Daniel Dr (CMP); biojava-dev at lists.open-bio.org
>>>>>Subject: Re: [Biojava-dev] Bug when reading FASTA file with many DNA
>>>>>Sequences
>>>>>
>>>>>Daniel
>>>>>
>>>>>You have two options. The first is to run java -Xmx2048m (the rest
>>>>>of your
>>>>>parameters) and the out of memory error will go away. I have a
>>>>>Helper method that will read the fasta file and lazy load when you
>>>>>request a sequence. If you call this method
>>>>>
>>>>>FastaReaderHelper.readFastaDNASequnece(File f, boolean
>>>lazySequenceLoad)
>>>>>you will be able to load the entire fasta file with minimal memory
>>>>>requirements.
>>>>>
>>>>>Even though your fasta file is X when we load it into memory each
>>>>>sequence position gets represented by a Java object so the memory
>>>>>footprint
>>>will
>>>>>be
>>>>>larger.
>>>>>
>>>>>Let me know if you don't have that particular method in the jars you
>>>are
>>>>>using. Not sure of the latest release on jars. If you look in the
>>>>>biojava3-genome module you will find examples of working with the
>>>>>DNA sequences to translate proteins etc assuming you have CDS
>>>>>features to map onto your sequences.
>>>>>
>>>>>Thanks
>>>>>
>>>>>Scooter
>>>>>
>>>>>On 2/9/11 7:13 AM, "Mapleson Daniel Dr (CMP)" <D.Mapleson at uea.ac.uk>
>>>>>wrote:
>>>>>
>>>>>>Hello,
>>>>>>
>>>>>>I'm trying to read a FASTA file that contains just over 4000 DNA
>>>>>>sequences and is around 270MB big. Each sequence starts like this:
>>>>>>">SequenceName" followed by a linefeed. The actual DNA sequence
>>>>>>data does contain a linefeed every 40 characters or so.
>>>>>>
>>>>>>I want to read in the data into a LinkedHashMap object, similar to
>>>the
>>>>>>example you specify in your cookbook:
>>>>>>
>>>>>>FileInputStream inStream = new FileInputStream(genomeFile);
>>>>>>FastaReader<DNASequence, NucleotideCompound> fastaReader = new
>>>>>>FastaReader<DNASequence, NucleotideCompound>(
>>>>>> inStream,
>>>>>> new GenericFastaHeaderParser<DNASequence,
>>>>>>NucleotideCompound>(),
>>>>>> new
>>>>>>DNASequenceCreator(AmbiguityDNACompoundSet.getDNACompoundSet()));
>>>>>>
>>>>>> try {
>>>>>> genomeData = fastaReader.process();
>>>>>> } catch (Exception ex) { }
>>>>>>
>>>>>>This works on some files but not the one containing the 4000
>>>sequences.
>>>>>>I get an exception generated by the JVM:
>>>>>>
>>>>>>Exception in thread "AWT-EventQueue-0" java.lang.OutOfMemoryError:
>>>>>>GC overhead limit exceeded
>>>>>> at
>>>>>>java.lang.AbstractStringBuilder.<init>(AbstractStringBuilder.java:4
>5)
>>>>>> at java.lang.StringBuilder.<init>(StringBuilder.java:80)
>>>>>> at
>>>>>>org.biojava3.core.sequence.io.FastaReader.process(FastaReader.java:
>>>>>>11
>>>1)
>>>>>> at
>>>>>>workbench.Process_Hits_Patman.openFastaFile(Process_Hits_Patman.jav
>a:
>>>41
>>>>>4)
>>>>>> at
>>>>>>workbench.Process_Hits_Patman.readFiles(Process_Hits_Patman.java:45
>1)
>>>>>> at workbench.MirCat.openFile(MirCat.java:283)
>>>>>> at
>>>>>>workbench.MainMDIWindow.openMenuItemActionPerformed(MainMDIWindow.j
>>>>>>av
>>>a:
>>>>>252
>>>>>>)
>>>>>> at
>workbench.MainMDIWindow.access$000(MainMDIWindow.java:25)
>>>>>> at
>>>>>>workbench.MainMDIWindow$1.actionPerformed(MainMDIWindow.java:140)
>>>>>> at
>>>>>>javax.swing.AbstractButton.fireActionPerformed(AbstractButton.java:
>>>>>>19
>>>95
>>>>>)
>>>>>> at
>>>>>>javax.swing.AbstractButton$Handler.actionPerformed(AbstractButton.j
>>>>>>av
>>>a:
>>>>>231
>>>>>>8)
>>>>>> at
>>>>>>javax.swing.DefaultButtonModel.fireActionPerformed(DefaultButtonMod
>>>>>>el
>>>.j
>>>>>ava
>>>>>>:387)
>>>>>> at
>>>>>>javax.swing.DefaultButtonModel.setPressed(DefaultButtonModel.java:2
>>>>>>42
>>>)
>>>>>> at
>>>javax.swing.AbstractButton.doClick(AbstractButton.java:357)
>>>>>> at
>>>>>>javax.swing.plaf.basic.BasicMenuItemUI.doClick(BasicMenuItemUI.java
>>>>>>:1
>>>22
>>>>>5)
>>>>>> at
>>>>>>javax.swing.plaf.basic.BasicMenuItemUI$Handler.mouseReleased(BasicM
>>>>>>en
>>>uI
>>>>>tem
>>>>>>UI.java:1266)
>>>>>> at
>java.awt.Component.processMouseEvent(Component.java:6263)
>>>>>> at
>>>>>javax.swing.JComponent.processMouseEvent(JComponent.java:3267)
>>>>>> at java.awt.Component.processEvent(Component.java:6028)
>>>>>> at java.awt.Container.processEvent(Container.java:2041)
>>>>>> at
>java.awt.Component.dispatchEventImpl(Component.java:4630)
>>>>>> at
>java.awt.Container.dispatchEventImpl(Container.java:2099)
>>>>>> at java.awt.Component.dispatchEvent(Component.java:4460)
>>>>>> at
>>>>>>java.awt.LightweightDispatcher.retargetMouseEvent(Container.java:45
>>>>>>74
>>>)
>>>>>> at
>>>>>>java.awt.LightweightDispatcher.processMouseEvent(Container.java:423
>8)
>>>>>> at
>>>>>>java.awt.LightweightDispatcher.dispatchEvent(Container.java:4168)
>>>>>> at
>java.awt.Container.dispatchEventImpl(Container.java:2085)
>>>>>> at java.awt.Window.dispatchEventImpl(Window.java:2475)
>>>>>> at java.awt.Component.dispatchEvent(Component.java:4460)
>>>>>> at java.awt.EventQueue.dispatchEvent(EventQueue.java:599)
>>>>>> at
>>>>>>java.awt.EventDispatchThread.pumpOneEventForFilters(EventDispatchTh
>>>>>>re
>>>ad
>>>>>.ja
>>>>>>va:269)
>>>>>> at
>>>>>>java.awt.EventDispatchThread.pumpEventsForFilter(EventDispatchThrea
>d.
>>>ja
>>>>>va:
>>>>>>184)
>>>>>>
>>>>>>The amount of memory on the system isn't an issue. I tried this on
>>>>>>a machine with 12GB of RAM. It seems to be an issue with the
>>>>>>garbage collector getting tired of deleting temporary objects!
>>>>>>Also I
>>>noticed
>>>>>>that although the file is less than 300MB large, the actual amount
>>>>>>of heap space used increases from 100MB to over 900MB when in
>>>>>>FastaReader.Process before the exception occurs.
>>>>>>
>>>>>>Unfortunately I can't share the FASTA file that is causing the
>>>problem.
>>>>>>
>>>>>>Would it be possible for you guys to look into this and either
>>>produce
>>>>>a
>>>>>>fix or suggest a workaround? Also do you think there is someway to
>>>>>>optimise the performance and memory usage of this process?
>>>>>>
>>>>>>Finally, I have a question about selectively loading sequences from
>>>>>>a FASTA file. The idea being to reduce memory usage. Is it
>>>possibility
>>>>>to
>>>>>>do this using biojava? i.e. given a DNA sequence name, only load
>>>that
>>>>>>sequence into memory? Or do we have to load the entire FASTA file
>>>into
>>>>>a
>>>>>>LinkedHashMap each time?
>>>>>>
>>>>>>Thanks in advance for your help on this one,
>>>>>>
>>>>>>Best regards,
>>>>>>Dr Daniel Mapleson (UEA)
>>>>>>
>>>>>>_______________________________________________
>>>>>>biojava-dev mailing list
>>>>>>biojava-dev at lists.open-bio.org
>>>>>>http://lists.open-bio.org/mailman/listinfo/biojava-dev
>>>>
>>
More information about the biojava-dev
mailing list